Disruptions to Qatar’s helium supply are exposing a structural blind spot in funds like VanEck Semiconductor ETF (NASDAQ:SMH) and iShares Semiconductor ETF (NASDAQ:SOXX), which are heavily loaded on chipmakers but lack exposure to the critical inputs that keep production running.

While oil disruptions typically dominate headlines, damage to Qatar’s Ras Laffan Industrial City, located above the world’s largest non-associated natural gas field, has brought attention to an invisible but equally vital input in semiconductor manufacturing.

According to a report early this week by UBS Global Wealth Management, cited by CNBC, Qatar accounts for roughly 30% of global helium supply. Industry estimates suggest the recent disruption could temporarily remove around 15% of global supply. The gap is due to the prior two years of oversupply, which is acting as insurance. Unlike other industrial inputs, helium is not easily substitutable or scalable in the short term, making the semiconductor supply chain particularly vulnerable.

Why Helium Matters More Than Investors Think

Helium is critical in chip manufacturing, especially in extreme ultraviolet (EUV) lithography for the latest nodes below 7nm.

More importantly, helium is a byproduct of liquefied natural gas (LNG) processing. This gives the supply chain a geographical dimension that is closely linked to the energy sector. This creates a dual dependency: gas-rich countries like the Gulf and the stability of global politics.

Even the leaders in the fabless chip business, such as Nvidia Corp (NASDAQ:NVDA), are not immune to this problem. They use companies like Taiwan Semiconductor Manufacturing Company Ltd (NYSE:TSM), which sources its industrial gases from around the world.

So, even a delay in the supply chain can affect GPU manufacturing. Given that AI infrastructure is expected to exceed current Street forecasts by $225 billion in 2027 and 2028 on the hyperscalers’ side alone (per Barclays), even short-lived supply shocks can ripple through data center buildouts and cloud deployments.

ETF Concentration Meets A New Kind Of Risk

Semiconductor ETFs have been some of the biggest winners in the AI bubble—but the structure of the ETFs may actually contribute to this risk.

For instance, SMH invests more than 30% of its assets in just two companies: Nvidia and TSMC, with additional exposure to Broadcom, Inc (NASDAQ:AVGO). Similarly, SOXX invests in around 30 stocks but remains heavily tilted toward a handful of AI-linked firms. Equal-weighting strategies like the SPDR S&P Semiconductor ETF (NYSE:XSD) also invest in the same global fabrication ecosystem.

These ETFs have generated 50-70% returns over the past year, driven by earnings growth fueled by the AI bubble. However, all these ETFs invest in the same technology companies that produce the chips and the equipment required for chip production—completely ignoring the upstream factors that actually enable the production process.

The Real Twist: Overexposed To Tech, Under-Hedged On Inputs

This is where the existing helium shock reframes the narrative.

The investors who are currently pouring money into semiconductor ETFs are, in a sense, making a one-sided bet that AI demand will keep rising. However, what is not being hedged against is a potential supply issue.

There is currently a lack of, or minimal, investment in industrial gases, LNG, and energy infrastructure companies, that sit upstream in the value chain.

What this means is that investors are heavily skewed towards companies that benefit from the AI trend, like Nvidia, but are not hedged against companies that enable semiconductor production.

If helium shortages persist, it is not a balanced situation where everyone is equally hurt: Upstream companies could benefit from tighter supply and rising prices, while semiconductor production could face a slowdown.

The helium disruption highlights that ETF diversification may not fully capture real-world dependencies. Semiconductor ETFs remain compelling long-term vehicles for AI exposure, but near-term volatility could rise if input constraints linger.

Image: Shutterstock